Method of processing an audio signal in a hearing device

The present application claims priority to EP patent application Ser. No. 23/185,992.7, filed Jul. 18, 2023, which is hereby incorporated by reference in its entirety.

Hearing devices may be used to improve the hearing capability or communication capability of a user, for instance by compensating a hearing loss of a hearing-impaired user, in which case the hearing device is commonly referred to as a hearing instrument such as a hearing aid, or hearing prosthesis. A hearing device may also be used to output sound based on an audio signal which may be communicated by a wire or wirelessly to the hearing device. A hearing device may also be used to reproduce a sound in a user's ear canal detected by an input transducer such as a microphone or a microphone array. The reproduced sound may be amplified to account for a hearing loss, such as in a hearing instrument, or may be output without accounting for a hearing loss, for instance to provide for a faithful reproduction of detected ambient sound and/or to add audio features of an augmented reality in the reproduced ambient sound, such as in a hearable. A hearing device may also provide for a situational enhancement of an acoustic scene, e.g. beamforming and/or active noise cancelling (ANC), with or without amplification of the reproduced sound. A hearing device may also be implemented as a hearing protection device, such as an earplug, configured to protect the user's hearing. Different types of hearing devices configured to be be worn at an ear include earbuds, earphones, hearables, and hearing instruments such as receiver-in-the-canal (RIC) hearing aids, behind-the-ear (BTE) hearing aids, in-the-ear (ITE) hearing aids, invisible-in-the-canal (IIC) hearing aids, completely-in-the-canal (CIC) hearing aids, cochlear implant systems configured to provide electrical stimulation representative of audio content to a user, a bimodal hearing system configured to provide both amplification and electrical stimulation representative of audio content to a user, or any other suitable hearing prostheses. A hearing system comprising two hearing devices configured to be worn at different ears of the user is sometimes also referred to as a binaural hearing device. A hearing system may also comprise a hearing device, e.g., a single monaural hearing device or a binaural hearing device, and a user device, e.g., a smartphone and/or a smartwatch, communicatively coupled to the hearing device.

Hearing devices are often employed in conjunction with communication devices, such as smartphones or tablets, for instance when listening to sound data processed by the communication device and/or during a phone conversation operated by the communication device. More recently, communication devices have been integrated with hearing devices such that the hearing devices at least partially comprise the functionality of those communication devices. A hearing system may comprise, for instance, a hearing device and a communication device.

In recent times, some hearing devices are also increasingly equipped with different sensor types. Traditionally, those sensors often include an input transducer to detect a sound, e.g., a sound detector such as a microphone or a microphone array. An amplified and/or signal processed version of the detected sound may then be outputted to the user by an output transducer, e.g., a receiver, loudspeaker, or electrodes to provide electrical stimulation representative of the outputted signal. In an effort to provide the user with even more information about himself and/or the ambient environment, various other sensor types are progressively implemented, in particular sensors which are not directly related to the sound reproduction and/or amplification function of the hearing device. Those sensors include inertial sensors, such as accelerometers, allowing to monitor the user's movements. Physiological sensors, such as optical sensors and bioelectric sensors, are mostly employed for monitoring the user's health.

Since the first digital hearing aid was created in the 1980s, hearing aids have been increasingly equipped with the capability to execute a wide variety of increasingly sophisticated audio processing algorithms intended not only to account for an individual hearing loss of a hearing impaired user but also to provide for a hearing enhancement in rather challenging environmental conditions and according to individual user preferences. Those increased signal processing capabilities, however, also come at a cost of increasingly demanding resources available in the hearing aid such as, e.g., processing power, memory availability and battery life. In this regard, hearing devices are more challenging than other devices due to a restricted amount of space available inside the car canal to accommodate increasingly sophisticated components.

In some cases or situations during usage of a hearing device, however, those sophisticated audio processing algorithms are not necessary or not even desirable to be applied. In particular, signal processing, e.g. deep neural network (DNN) based signal processing, for the purpose of generating audiological user benefit such as, e.g., improved clarity of speech can come with side effects and/or downsides such as an increased power consumption, thus reduced battery life-time, and/or signal processing artefacts and/or unnatural sound perception which are limiting the acceptance of the signal processing by the user and thus limiting its benefit. E.g., in some situations, the user may prefer a longer battery life of the hearing device as compared to an elaborate but also increasingly complex signal processing technique. Further, at least in some situations, the user may also dislike negative side effects caused by the more complex signal processing which may include, e.g., an increased latency and or more pronounced artefacts in the sound reproduced by the hearing device.

Achieving the best overall balance involves making a trade-off between the benefits and the downsides. The trade-off cannot be fully determined by a priori factors such as hearing loss, but varies, e.g., with a user preference, listening intention, life-style/habits and a current situation the user is in. Accordingly, the balance should be variable and under control of the user and/or health care professional (HCP), allowing for a best trade-off and optimally also meeting the user's needs. Typically, the trade-off is experienced/perceived by the user in a real-life situation and/or in a particular use-case and adjusted through a proper interface such as, e.g., an app or through gestures which may also be performed directly on the hearing device. Therefore, an adaptability and/or selection of a momentarily performed audio processing, which may depend on a current situation and/or user preference, would be highly desirable. In particular, an intelligent system may learn and/or estimate the end-user preferred or intended balance and gradually relax the user from taking corrective action while still providing best results.

The disclosure relates to method of operating a hearing device configured to be worn at an ear of a user.

It is a feature of the present disclosure to avoid at least one of the above mentioned disadvantages and to apply an audio processing algorithm which has been selected from a plurality of available algorithms in an optimized way, e.g., on demand of the user and/or commensurate with a current hearing situation. It is another feature to provide for an audio processing in different situations which takes into account the positive and negative side effects of different audio processing algorithms available in the hearing device. It is yet another feature to equip a hearing device with a capability to apply such an optimally selected audio processing algorithm on an input audio signal, in particular in an automated and/or user-selected way. It is a further feature to allow the hearing device to better manage its available resources when it comes to performing a suitable audio processing.

receiving an audio signal; processing the audio signal by at least one audio processing algorithm to generate a processed audio signal; and outputting, by an output transducer included in the hearing device, an output audio signal based on the processed audio signal so as to stimulate the user's hearing, wherein the method further comprises providing different audio processing algorithms each configured to be applied on the audio signal and associated with a performance index indicative of a performance of the audio processing algorithm when applied on the audio signal; determining a target index relative to the performance index, the target index indicative of a target performance of said processing of the audio signal; selecting, depending on the target index, at least one of the processing algorithms; and applying the selected processing algorithm on the audio signal. Accordingly, the present disclosure proposes a method of operating a hearing device configured be worn at an ear of a user, the method comprising

In this way, by selecting an audio processing algorithm with a suitable performance index based on the target index, the audio processing can be advantageously adapted to a current hearing situation and/or user requirement. In particular, when applying the audio processing algorithm with the suitable performance index, available resources can be used sparingly and/or negative side effects of the audio processing can be circumvented by also reaching a desired goal of the audio processing. By associating the different processing algorithms with a corresponding performance index, the algorithms can thus be scaled in accordance with an expected processing performance. For instance, a rather memory intensive, power costly and time consuming operation involving a deep neural network (DNN) may then be replaced in favor of another audio processing algorithm which may be more efficient and still allow to reach the desired signal processing goal or may even be suitable to exceed the DNN in at least some aspects of the signal processing. E.g., the target index may be indicative of a desired performance when one or more of the audio processing algorithms are applied on the audio signal.

Independently, the present disclosure also proposes a non-transitory computer-readable medium storing instructions that, when executed by a processor, cause a hearing device to perform operations of the method.

a processor configured to process the audio signal by at least one audio processing algorithm to generate a processed audio signal; and provide different audio processing algorithms each configured to be applied on the audio signal and associated with a performance index indicative of a performance of the audio processing algorithm when applied on the audio signal; determine a target index relative to the performance index, the target index indicative of a target performance of said processing of the audio signal; select, depending on the target index, at least one of the processing algorithms; and apply the selected processing algorithm on the audio signal. an output transducer configured to output an output audio signal based on the processed audio signal so as to stimulate the user's hearing, wherein the processor is further configured to Independently, the present disclosure also proposes a hearing device configured be worn at an ear of a user, the hearing device comprising an input transducer configured to provide an audio signal indicative of a sound detected in the environment of the user;

the first hearing device comprising a first input transducer configured to provide a first audio signal indicative of sound detected in the environment of the user, and the second hearing device comprising a second input transducer configured to provide a second audio signal indicative of sound detected in the environment of the user; the hearing system further comprising a processor configured to process the first and second audio signal by at least one audio processing algorithm to generate a processed audio signal; and provide different audio processing algorithms each configured to be applied on the first and second audio signal and associated with a performance index indicative of a performance of the audio processing algorithm when applied on the first and second audio signal; determine a target index relative to the performance index, the target index indicative of a target performance of said processing of the audio signal; select, depending on the target index, at least one of the processing algorithms; and apply the selected processing algorithm on the first and second audio signal. the first hearing device further comprising a first output transducer configured to output a first output audio signal based on the processed audio signal, and the second hearing device further comprising a second output transducer configured to output a second output audio signal based on the processed audio signal so as to stimulate the user's hearing, wherein the processor is further configured to Independently, the present disclosure also proposes a hearing system comprising a first hearing device and a second hearing device each configured be worn at a different ear of a user,

Subsequently, additional features of some implementations of the method of operating a hearing device and/or the computer-readable medium and/or the hearing device are described. Each of those features can be provided solely or in combination with at least another feature. The features can be correspondingly provided in some implementations of the method and/or the hearing device.

a dimension indicative of an impact of the audio processing algorithm on resources available in the hearing device; and/or a dimension indicative of an enhancement of the hearing perception of the user by the processing of the audio signal; and/or a dimension indicative of an adverse effect of the processing of the audio signal for the hearing perception of the user. In some implementations, the performance index has at least one dimension comprising

a power consumption of the algorithm, e.g., relative to a life of a battery included in the hearing device; a computational load of executing the algorithm; a memory requirement of the algorithm; and a communication bandwidth required to execute the algorithm in a distributed processor comprising at least two processing units communicating with each other. In some implementations, the impact of the audio processing algorithm on available resources comprises at least one of

a measure of a clarity of sound encoded in the audio signal; a measure of an understandability of a speech encoded in the audio signal; a measure of a listening effort needed for understanding information encoded in the audio signal; a measure of a comfort when listening to sound encoded in the audio signal; a measure of a naturalness of sound encoded in the audio signal; a measure of a spatial perceptibility of sound encoded in the audio signal; and a measure of a quality of sound encoded in the audio signal. In some implementations, the enhancement of the hearing perception of the user comprises at least one of

a level of artefacts in the processed audio signal; a level of distortions of sound encoded in the processed audio signal; and a level of a latency for outputting the output audio signal based on the processed audio signal. In some implementations, the adverse effect of the processing comprises at least one of

receiving, from a user interface, a user command indicative of the target index; evaluating the audio signal, wherein the target index is determined based on the evaluated audio signal; receiving, from a sensor included in the hearing device, sensor data, wherein the target index is determined based on the sensor data; and acquiring information about resources available in the hearing device, wherein the target index is determined based on the information. In some implementations, the determining the target index comprises at least one of

In some implementations, the target index is indicative of an applicability of the audio processing algorithms for a processing of the audio signal. The target index may then also be referred to as an applicability index. E.g., the target index may constrain the applicability of the audio processing algorithms. The target index may then also be referred to as a constraint index.

classifying the audio signal by attributing at least one class from a plurality of predetermined classes to the audio signal, wherein the target index is determined depending on the class attributed to the audio signal. In some implementations, the method further comprises

In some implementations, the user command is indicative of a value desired by the user of the performance index. E.g., the user command may be indicative of a value desired by the user of the performance index in at least one of said dimensions.

an enhancement of a speech content of a single talker in the audio signal; an enhancement of a speech content of a plurality of talkers in the audio signal; a reproduction of sound emitted by an acoustic object in the environment of the user encoded in the audio signal; a reproduction of sound emitted by a plurality of acoustic objects in the environment of the user encoded in the audio signal; a reduction and/or cancelling of noise and/or reverberations in the audio signal; a preservation of acoustic cues contained in the audio signal; a suppression of noise in the audio signal; an improvement of a signal to noise ratio (SNR) in the audio signal; a spatial resolution of sound encoded in the audio signal depending on a direction of arrival (DOA) of the sound and/or depending on a location of at least one acoustic object emitting the sound in the environment of the user; a directivity of an audio content in the audio signal provided by a beamforming or a preservation of an omnidirectional audio content in the audio signal; an amplification of sound encoded in the audio signal adapted to an individual hearing loss of the user; and an enhancement of music content in the audio signal.E.g., the performance index may be differing in one or more of said dimensions. In some implementations, the different audio processing algorithms comprise at least two audio processing algorithms configured to provide for the same signal processing goal which are associated with a differing performance index, wherein the signal processing goal comprises at least one of

In some implementations, the different audio processing algorithms comprise a first set of audio processing algorithms and a second set of audio processing algorithms, wherein at least one of the audio processing algorithms of the first set and at least one of the audio processing algorithms of the second set are configured to provide for the same signal processing goal and are associated with a differing performance index. E.g., the performance index may be differing in one or more of said dimensions.

In some implementations, depending on the target index, at least two of the audio processing algorithms of the first set or the second set are selected to be applied in a sequence and/or in parallel on the audio signal to generate the processed audio signal.

In some implementations, at least one of the audio processing algorithms is included in the first set and in the second set.

In some implementations, the first set and the second set are associated with a performance index indicative of the performance index of each of the audio processing algorithms included in the set.

In some implementations, the audio processing algorithms comprise at least one neural network (NN).

In some implementations, the NN comprises an encoder part configured to encode the audio signal, and a decoder part configured to decode the encoded audio signal.

In some implementations, the different audio processing algorithms comprise a first NN comprising the encoder part and a first decoder part, and a second NN comprising the encoder part and a second decoder part differing from the first decoder part, wherein the first NN and the second NN are associated with a differing performance index. E.g., the performance index may be differing in one or more of said dimensions.

In some implementations, the first set of audio processing algorithms comprises the first NN, and the second set of audio processing algorithms comprises the second NN.

In some implementations, the audio signal is indicative of a sound in the ambient environment of the user. In some implementations, the audio signal is received from an input transducer, e.g., a microphone or a microphone array, included in the hearing device. In some implementations, the audio signal is received by an audio signal receiver included in the hearing device, e.g., via radio frequency (RF) communication. In some implementations, the audio signal is received from a remote microphone, e.g., a table microphone and/or a clip-on microphone.

1 FIG. 110 110 110 110 illustrates an exemplary hearing deviceconfigured to be worn at an car of a user. Hearing devicemay be implemented by any type of hearing device configured to enable or enhance hearing or a listening experience of a user wearing hearing device. For example, hearing devicemay be implemented by a hearing aid configured to provide an amplified version of audio content to a user, a sound processor included in a cochlear implant system configured to provide electrical stimulation representative of audio content to a user, a sound processor included in a bimodal hearing system configured to provide both amplification and electrical stimulation representative of audio content to a user, or any other suitable hearing prosthesis, or an earbud or an earphone or a hearable.

110 Different types of hearing devicecan also be distinguished by the position at which they are worn at the ear. Some hearing devices, such as behind-the-ear (BTE) hearing aids and receiver-in-the-canal (RIC) hearing aids, typically comprise an earpiece configured to be at least partially inserted into an ear canal of the ear, and an additional housing configured to be worn at a wearing position outside the ear canal, in particular behind the ear of the user. Some other hearing devices, as for instance earbuds, earphones, hearables, in-the-ear (ITE) hearing aids, invisible-in-the-canal (IIC) hearing aids, and completely-in-the-canal (CIC) hearing aids, commonly comprise such an earpiece to be worn at least partially inside the ear canal without an additional housing for wearing at the different ear position.

110 112 113 114 117 114 115 116 110 119 110 118 112 110 115 117 As shown, hearing deviceincludes a processorcommunicatively coupled to a memory, an audio input unit, and an output transducer. Audio input unitmay comprise at least one input transducerand/or an audio signal receiverconfigured to provide an input audio signal. Hearing devicemay further include a communication port. Hearing devicemay further include a sensor unitcommunicatively coupled to processor. Hearing devicemay include additional or alternative components as may serve a particular implementation. Input transducermay be implemented by any suitable device configured to detect sound in the environment of the user and to provide an input audio signal indicative of the detected sound, e.g., a microphone or a microphone array. Output transducermay be implemented by any suitable audio transducer configured to output an output audio signal to the user, for instance a receiver of a hearing aid, an output electrode of a cochlear implant system, or a loudspeaker of an earbud.

112 114 115 116 112 117 Processoris configured to receive, from audio input unit, an input audio signal. E.g., when the audio signal is received from input transducer, the audio signal may be indicative of a sound detected in the environment of the user and/or, when the audio signal is received from audio signal receiver, the audio signal may be indicative of a sound provided from a remote audio source such as, e.g., a remote microphone and/or an audio streaming server. Processoris further configured to process the audio signal by at least one audio processing algorithm to generate a processed audio signal; and to control output transducerto output an output audio signal based on the processed audio signal so as to stimulate the user's hearing.

112 112 112 Processoris also configured to provide different audio processing algorithms each configured to be applied on the audio signal and indicative of a performance of the audio processing algorithm when applied on the audio signal. Processoris further configured to determine a target index relative to the performance index, to select, depending on the target index, at least one of the processing algorithms, and to apply the selected processing algorithm on the audio signal. These and other operations, which may be performed by processor, are described in more detail in the description that follows.

113 112 112 113 112 115 115 116 113 112 112 Memorymay be implemented by any suitable type of storage medium and is configured to maintain, e.g. store, data controlled by processor, in particular data generated, accessed, modified and/or otherwise used by processor. For example, memorymay be configured to store instructions used by processorto process the input audio signal received from input transducer, e.g., audio processing instructions in the form of one or more audio processing algorithms. The audio processing algorithms may comprise different audio processing instructions of processing the input audio signal received from input transducerand/or audio signal receiver. For instance, the audio processing algorithms may provide for at least one of a gain model (GM) defining an amplification characteristic, a noise cancelling (NC) algorithm, a wind noise cancelling (WNC) algorithm, a reverberation cancelling (RevC) algorithm, a feedback cancelling (FC) algorithm, a speech enhancement (SE) algorithm, a gain compression (GC) algorithm, a noise cleaning algorithm, a binaural synchronization (BS) algorithm, a beamforming (BF) algorithm, in particular static and/or adaptive beamforming, and/or the like. Further examples of audio processing algorithms, which may be stored in memoryand/or applied by processor, are described in the following description. A plurality of the audio processing algorithms may be executed by processorin a sequence and/or in parallel to generate a processed audio signal.

113 112 115 116 As another example, memorymay be configured to store instructions used by processorto classify the input audio signal received from input transducerand/or audio signal receiverby attributing at least one class from a plurality of predetermined sound classes to the input audio signal. Exemplary classes may include, but are not limited to, low ambient noise, high ambient noise, traffic noise, music, machine noise, babble noise, public area noise, background noise, speech, nonspeech, speech in quiet, speech in babble, speech in noise, speech from the user, speech from a significant other, background speech, speech from multiple sources, quiet indoor, quiet outdoor, speech in a car, speech in traffic, speech in a reverberating environment, speech in wind noise, speech in a lounge, car noise, applause, music, e.g. classical music, and/or the like. In some instances, the different audio processing instructions can be associated with different classes.

113 113 113 Memorymay comprise a non-volatile memory from which the maintained data may be retrieved even after having been power cycled, for instance a flash memory and/or a read only memory (ROM) chip such as an electrically erasable programmable ROM (EEPROM). A non-transitory computer-readable medium may thus be implemented by memory. Memorymay further comprise a volatile memory, for instance a static or dynamic random access memory (RAM).

110 119 119 110 119 As illustrated, hearing devicemay further comprise a communication port. Communication portmay be implemented by any suitable data transmitter and/or data receiver and/or data transducer configured to exchange data with another device. For instance, the other device may be another hearing device configured to be worn at the other car of the user than hearing deviceand/or a communication device such as a smartphone, smartwatch, tablet and/or the like. Communication portmay be configured for wired and/or wireless data communication. For instance, data may be communicated in accordance with a Bluetooth™ protocol and/or by any other type of radio frequency (RF) communication.

110 116 116 116 116 116 As illustrated, hearing devicemay further comprise an audio signal receiver. Audio signal receivermay be implemented by any suitable data receiver and/or data transducer configured to receive an input audio signal from a remote audio source. For instance, the remote audio source may be a wireless microphone, such as a table microphone, a clip-on microphone and/or the like, and/or a portable device, such as a smartphone, smartwatch, tablet and/or the like, and/or any another data transceiver configured to transmit the input audio signal to audio signal receiver. E.g., the remote audio source may be a streaming source configured for streaming the input audio signal to audio signal receiver. Audio signal receivermay be configured for wired and/or wireless data reception of the input audio signal. For instance, the input audio signal may be received in accordance with a Bluetooth™ protocol and/or by any other type of radio frequency (RF) communication.

110 118 112 115 118 2 FIG. As illustrated, hearing devicemay comprise a sensor unitcomprising at least one further sensor communicatively coupled to processorin addition to input transducer. Some examples of a sensor which may be implemented in sensor unitare illustrated in.

2 FIG. 118 115 130 131 132 118 133 134 135 133 133 134 As illustrated in, sensor unitmay include at least one environmental sensor configured to provide environmental data indicative of a property of the environment of the user in addition to input transducer, for example an optical sensorconfigured to detect light in the environment and/or a barometric sensorand/or an ambient temperature sensor. Sensor unitmay include at least one physiological sensor configured to provide physiological data indicative of a physiological property of the user, for example an optical sensorand/or a bioelectric sensorand/or a body temperature sensor. Optical sensormay be configured to emit the light at a wavelength absorbable by an analyte contained in blood such that the physiological sensor data comprises information about the blood flowing through tissue at the car. E.g., optical sensorcan be configured as a photoplethysmography (PPG) sensor such that the physiological sensor data comprises PPG data, e.g. a PPG waveform. Bioelectric sensormay be implemented as a skin impedance sensor and/or an electrocardiogram (ECG) sensor and/or an electroencephalogram (EEG) sensor and/or an electrooculography (EOG) sensor.

118 136 118 137 110 118 138 118 139 115 131 137 138 139 119 115 131 137 118 Sensor unitmay include a movement sensorconfigured to provide movement data indicative of a movement of the user, for example an accelerometer and/or a gyroscope and/or a magnetometer. Sensor unitmay include a user interfaceconfigured to provide interaction data indicative of an interaction of the user with hearing device, e.g., a touch sensor and/or a push button. Sensor unitmay include at least one location sensorconfigured to provide location data indicative of a current location of the user, for instance a GPS sensor. Sensor unitmay include at least one clockconfigured to provide time data indicative of a current time. Context data may be defined as data indicative of a local and/or temporal context of the data provided by other sensors,-. Context data may comprise the location data and/or the time data provided by location sensorand/or clock. Context data may also be received from an external device via communication port, e.g., from a communication device. E.g., one or more of sensors,-may then be included in the communication device. Sensor unitmay include further sensors providing sensor data indicative of a property of the user and/or the environment and/or the context.

3 FIG. 110 210 210 220 240 220 221 221 112 115 116 220 227 240 241 241 117 240 145 220 240 251 112 117 145 240 251 252 253 221 241 130 139 220 240 illustrates an exemplary implementation of hearing deviceas a RIC hearing aid. RIC hearing aidcomprises a BTE partconfigured to be worn at an car at a wearing position behind the car, and an ITE partconfigured to be worn at the car at a wearing position at least partially inside an car canal of the car. BTE partcomprises a BTE housingconfigured to be worn behind the car. BTE housingaccommodates processorcommunicatively coupled to input transducerand audio signal receiver. BTE partfurther includes a batteryas a power source. ITE partis an earpiece comprising an ITE housingat least partially insertable in the car canal. ITE housingaccommodates output transducer. ITE partmay further include an in-the-car input transducer, e.g., an car canal microphone, configured to detect sound inside the car canal and to provide an in-the-car audio signal indicative of the detected sound. BTE partand ITE partare interconnected by a cable. Processoris communicatively coupled to output transducerand to in-the-car input transducerof ITE partvia cableand cable connectors,provided at BTE housingand ITE housing. In some implementations, at least one of sensors-is included in BTE partand/or ITE part.

4 FIG. 310 110 120 310 120 110 110 120 210 120 122 123 127 124 115 116 120 129 illustrates an exemplary hearing systemcomprising first hearing deviceconfigured to be worn at a first car of the user, and a second hearing deviceconfigured to be worn at a second car of the user. Hearing systemmay also be denoted as a binaural hearing device. Second hearing devicemay be implemented corresponding to first hearing device. E.g., first hearing deviceand second hearing devicemay each be implemented corresponding to RIC hearing aiddescribed above. As shown, second hearing deviceincludes a processorcommunicatively coupled to a memory, an output transducer, an audio input unit, which may comprise at least one input transducer corresponding to input transducerand/or at least one audio signal receiver corresponding to audio signal receiver. Second hearing devicefurther includes a communication port.

112 110 122 120 119 129 318 112 110 122 112 122 112 122 112 122 112 122 310 112 122 119 129 Processorof first hearing deviceand processorof second hearing devicecan be communicatively coupled by communication ports,via a communication link. In this way, processorof first hearing devicemay form a first processing unit and processorof second hearing device may form a second processing unit of a processor comprising the first processing unitand the second processing unit. For instance, processor,may then be implemented as a distributed processing system of first processing unitand second processing unitand/or may operate in a master-slave configuration of first processing unitand second processing unit. Hearing systemmay further comprise a portable device, e.g., a communication device such as a smartphone, smartwatch, tablet and/or the like. The portable device, in particular a processor included in the portable device, may also be communicatively coupled to processors,, e.g., via communication ports,.

5 FIG. 505 112 122 505 502 503 504 502 505 110 120 210 503 504 505 505 502 504 illustrates an abstract view of different audio processing algorithmswhich may be executed by processorand/or processorto be applied on an audio signal. As illustrated, audio processing algorithmscan be organized in at least one dimension,,. The dimensions can include a dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,, a dimensionindicative of an enhancement of the hearing perception of the user by the processing of the audio signal, and a dimensionindicative of an adverse effect of the processing of the audio signal for the hearing perception of the user. A performance index associated with each of the different audio processing algorithms, which may be indicative of a performance of the respective audio processing algorithmwhen applied on the audio signal, may be defined as an index having at least one of dimensions-.

502 505 110 120 210 227 110 120 210 112 122 505 112 122 505 112 122 119 129 To illustrate, dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,may be indicative of at least one of a power consumption of the respective algorithm, thus affecting a life of batteryincluded in hearing device,,, a computational load of executing the algorithm, e.g., with regard to an available processing power of any of processor,, a memory requirement of the algorithm, e.g., of available volatile and/or non-volatile memory which may be used or accessed during execution of the respective algorithmby processor,, and a communication bandwidth required to execute the respective algorithm, e.g., in a distributed processor comprising at least two processing units,communicating with each other via communication ports,.

503 505 Dimensionindicative of an enhancement of the hearing perception of the user by the processing of the audio signal by the respective algorithmmay be indicative of at least one of a measure of a clarity of sound encoded in the audio signal; a measure of an understandability of a speech encoded in the audio signal; a measure of a listening effort needed for understanding information encoded in the audio signal; a measure of a comfort when listening to sound encoded in the audio signal; a measure of a naturalness of sound encoded in the audio signal; a measure of a spatial perceptibility of sound encoded in the audio signal; and a measure of a quality of sound encoded in the audio signal.

504 505 505 505 505 Dimensionindicative of an adverse effect of the processing of the audio signal by the respective algorithmfor the hearing perception of the user may be indicative of at least one of a level of artefacts in the processed audio signal, which may be caused by the processing by the respective algorithm; a level of distortions of sound encoded in the audio signal, which may be caused by the processing by the respective algorithm; and a level of a latency for outputting the output audio signal based on the processed audio signal, which may be caused by the processing by the respective algorithm.

6 FIG. 601 110 210 310 601 602 115 125 604 116 126 602 603 604 605 501 614 117 127 601 616 618 137 130 136 138 139 118 601 614 614 110 210 illustrates a functional block diagram of an exemplary audio signal processing arrangementthat may be implemented by hearing device,and/or hearing system. Arrangementcomprises at least one input transducer, which may be implemented by input transducer,, and/or at least one audio signal receiver, which may be implemented by audio signal receiver,. The audio signal provided by input transducermay be an analog signal. The analog signal may be converted into a digital signal by an analog-to-digital converter (ADC). The audio signal provided by audio signal receivermay be an encoded signal. The encoded signal may be decoded into a decoded signal by a decoder (DEC). Arrangementfurther comprises at least one output transducer, which may be implemented by output transducer,. Arrangementmay further comprise at least one user input unitand/or sensor unit, which may be implemented by user interfaceand/or at least one of sensors-,,included in sensor unit. Arrangementfurther comprises a hearing device management module. Hearing device management modulecan be configured to acquire information about resources currently available in hearing device,.

601 617 617 602 604 618 136 Arrangementmay further comprise a classifier. Classifiercan be configured to attribute at least one class to the audio signal provided by input transducerand/or audio signal receiverand/or at least one class to sensor data provided by sensor unit. E.g., when the class is attributed to the audio signal, the class attributed to the audio signal may include at least one of low ambient noise, high ambient noise, traffic noise, music, machine noise, babble noise, public area noise, background noise, speech, nonspeech, speech in quiet, speech in babble, speech in noise, speech from the user, speech from a significant other, background speech, speech from multiple sources, quiet indoor, quiet outdoor, speech in a car, speech in traffic, speech in a reverberating environment, speech in wind noise, speech in a lounge, car noise, applause, music, e.g. classical music, and/or the like is attributed to the audio signal. E.g., when the class is attributed to the sensor data, which may be provided by movement sensor, attributed to the movement data may comprise at least one of the user walking, running, standing, the user turning his head, and the user falling to the ground.

601 623 625 627 629 623 625 627 629 112 122 112 110 122 120 627 113 110 123 120 Arrangementfurther comprises a target index determination module, an audio processing algorithm selection module, an audio processing algorithm storage module, and an audio processing module. Modules,,,may be executed by at least one processor,, e.g., by a processing unit including processorof first hearing deviceand/or processorof second hearing device. Additionally or alternatively, audio processing algorithm storage modulemay be provided by at least one memory, e.g., by memoryof first hearing deviceand/or memoryof second hearing device.

602 603 604 605 629 629 614 615 614 As illustrated, the audio signal provided by input transducer, after it has been converted into a digital signal by analog-to-digital converter, and/or the audio signal provided by audio signal receiver, after it has been decoded by decoder, can be received by audio processing module. Audio processing moduleis configured to process the audio signal by applying one or more audio processing algorithms on the audio signal to generate a processed audio signal. In a case in which a plurality of audio processing algorithms are applied on the audio signal, the audio processing algorithms may be executed in a sequence and/or in parallel to generate the processed audio signal. Based on the processed audio signal, an output audio signal can be output by output transducerso as to stimulate the user's hearing. To this end, the processed audio signal may be converted into an analog signal by a digital-to-analog converter (DAC)before providing the processed audio signal to output transducer.

627 629 505 505 502 504 627 505 113 123 627 112 122 Audio processing algorithm storage moduleis configured to store a plurality of different audio processing algorithms. Each of the audio processing algorithms is configured to be applied on the audio signal by audio processing module. Further, each of the audio processing algorithms is associated with a performance index indicative of a performance of the audio processing algorithm when applied on the audio signal. E.g., the different audio processing algorithms may include audio processing algorithmsdescribed above. Each of audio processing algorithmsmay then be associated with a performance index having at least one of dimensions-. For instance, audio processing algorithm storage modulemay comprise a volatile and/or involatile memory to store audio processing algorithms, e.g., memory,. Additionally or alternatively, audio processing algorithm storage modulemay comprise an internal memory, e.g., a volatile memory, included in processor,.

505 627 629 505 For example, audio processing algorithmswhich may be stored by audio processing algorithm storage moduleand/or applied on the audio signal by audio processing module, may comprise at least one of a gain model (GM), which may define an amplification characteristic, e.g., to compensate for an individual hearing loss of the user; a noise cancelling (NC) algorithm; a wind noise cancelling (WNC) algorithm; a reverberation cancelling (RevC) algorithm; a feedback cancelling (FC) algorithm; a speech enhancement (SE) algorithm; an impulse noise cancelling (INC) algorithm; an acoustic object separation (AOS) algorithm; a binaural synchronization (BS) algorithm; and a beamforming (BF) algorithm, in particular adapted for static and/or adaptive beamforming. Further examples of audio processing algorithmsare described in the following description.

616 617 The gain model (GM) may comprise a gain compression (GC) algorithm which may be configured to provide for an amplification characteristic of the input audio signal which may depend on a loudness level of the audio content in the input audio signal. E.g., the amplification may be decreased, e.g., limited, for audio content having a higher signal level and/or the amplification may be increased, e.g., expanded, for audio content having a lower signal level. An operation of the gain compression (GC) algorithm may also be adjusted depending on a user command received from user interfaceand/or when classifierattributes at least one class to the audio signal. The gain model (GM) may also comprise a frequency compression (FreqC) algorithm which may be configured to provide for an amplification characteristic of the input audio signal which may depend on a frequency of the audio content in the input audio signal, e.g., to provide for audio content detected at higher frequencies an amplification shifted to a lower frequency band.

502 504 502 505 110 120 210 503 504 505 Some examples of different GC algorithms may comprise a low delay compression ratio (CR)/gain algorithm, which may provide for the gain compression in a more basic manner, and/or an advanced CR/gain algorithm, which may provide for the gain compression in a more sophisticated manner. The different GC algorithms may thus be each associated with a performance index differing in at least one of dimensions-. To illustrate, the different GC algorithms may differ in dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,due to an increasing power consumption and/or computational load and/or memory requirement and/or communication bandwidth when executing the advanced CR/gain algorithm in place of the low delay compression ratio (CR)/gain algorithm. Further, the different NC algorithms may differ in dimensionindicative of an enhancement of the hearing perception of the user achieved by the processing of the audio signal due to an improved gain compression and/or improved quality of the audio signal when executing the advanced CR/gain algorithm in place of the low delay compression ratio (CR)/gain algorithm, which may affect the listening effort and/or comfort and/or naturalness and/or other quality of the sound encoded in the audio signal. Further, the different NC algorithms may differ in dimensionindicative of an adverse effect of the processing of the audio signal by the respective algorithmfor the hearing perception of the user due to an increasing level of artefacts and/or distortions and/or latency when executing the advanced CR/gain algorithm in place of the low delay compression ratio (CR)/gain algorithm. Other examples of different GM algorithms may comprise an expansion algorithm, which may provide for a level expansion and/or a frequency expansion in the audio signal, and/or a maximum power output (MPO) algorithm, which may control a maximum power output.

602 604 629 617 616 618 The noise cancelling (NC) algorithm can be configured to provide for a cancelling and/or suppression and/or cleaning of noise contained in the audio signal. In some instances, the NC algorithm may be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver. For instance, the NC algorithm may be applied on the audio signal by audio processing modulewhen classifierattributes at least one class such as low ambient noise, high ambient noise, traffic noise, noise, babble noise, public area noise, background noise, speech, nonspeech, speech in quiet, speech in noise, speech in loud noise, speech in traffic, car noise, applause, and/or the like to the audio signal. A corresponding signal processing goal of the cancelling and/or suppression and/or cleaning of wind noise in the audio signal may thus be predicted based on a class attributed to the audio signal. The NC algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

Some examples of different noise cancelling (NC) algorithms may comprise a low delay NC algorithm, which may provide for the noise cancelling in a non-spatially resolved manner, and/or a traditional/hybrid NC algorithm, which may provide for a non-spatially resolved noise cancelling in a more sophisticated manner, and/or a directional NC algorithm, which may provide for the noise cancelling in a specific direction or location relative to the user, e.g., toward the front of the user, which may also be referred to as a front NC algorithm, and/or a denoising algorithm implemented by a neural network (NN), e.g., a deep neural network (DNN), which may also provide for the noise cancelling in a non-spatial manner.

502 504 502 505 110 120 210 503 504 505 The different NC algorithms may thus be each associated with a performance index differing in at least one of dimensions-. To illustrate, the different NC algorithms may differ in dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,due to an increasing power consumption and/or computational load and/or memory requirement and/or communication bandwidth when executing the directional NC algorithm in place of the low delay NC algorithm and/or when executing the denoising algorithm implemented by a NN in place of the low delay NC algorithm and/or the directional NC algorithm. Further, the different NC algorithms may differ in dimensionindicative of an enhancement of the hearing perception of the user achieved by the processing of the audio signal due to an increasing amount of the noise cancelling and/or improved quality of the audio signal when executing the directional NC algorithm in place of the low delay NC algorithm and/or when executing the denoising algorithm implemented as a NN in place of the low delay NC algorithm and/or the directional NC algorithm, which may affect the listening effort and/or comfort and/or naturalness and/or other quality of the sound encoded in the audio signal. Further, the different NC algorithms may differ in dimensionindicative of an adverse effect of the processing of the audio signal by the respective algorithmfor the hearing perception of the user due to an increasing level of artefacts and/or distortions and/or latency when executing the directional NC algorithm in place of the low delay NC algorithm and/or when executing the denoising algorithm implemented by a NN in place of the low delay NC algorithm and/or the directional NC algorithm.

602 604 629 617 616 618 The wind noise cancelling (WNC) algorithm can be configured to provide for a cancelling and/or suppression and/or cleaning of wind noise contained in the audio signal. In some instances, the WNC algorithm may be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver. For instance, the WNC algorithm may be applied on the audio signal by audio processing modulewhen classifierattributes at least one class such as wind noise to the audio signal. A corresponding signal processing goal of the cancelling and/or suppression and/or cleaning of wind noise in the audio signal may thus be predicted based on a class attributed to the audio signal. The WNC algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

602 604 629 617 616 618 The reverberation cancelling (RevC) algorithm can be configured to provide for a cancelling and/or suppression and/or cleaning of reverberations contained in the audio signal. In some instances, the RevC algorithm may be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver. For instance, the RevC algorithm may be applied on the audio signal by audio processing modulewhen classifierattributes at least one class such as reverberations and/or speech in a reverberating environment and/or the like to the audio signal. A corresponding signal processing goal of the cancelling and/or suppression and/or cleaning of reverberations in the input audio signal may thus be predicted based on a class attributed to the audio signal. The RevC algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

629 602 604 616 618 The feedback cancelling (FC) algorithm can be configured to provide for a cancelling and/or suppression and/or cleaning of feedback contained in the audio signal. For instance, the feedback cancelling (FC) algorithm may be executed by default by audio processing module, e.g., to account for a previously known signal processing goal to compensate for the feedback which may be present in the audio signal. The FC algorithm may also be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver, e.g., when feedback has been determined to be contained in the audio signal. The FC algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

502 504 502 505 110 120 210 503 504 505 Some examples of different FC algorithms may comprise a low delay FC management algorithm, and an FC algorithm providing for frequency shift and/or phase cancelling. The different FC algorithms may thus be each associated with a performance index differing in at least one of dimensions-. To illustrate, the different FC algorithms may differ in dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,due to an increasing power consumption and/or computational load and/or memory requirement and/or communication bandwidth when executing the FC algorithm providing for frequency shift and/or phase cancelling in place of the low delay FC management algorithm. Further, the different FC algorithms may differ in dimensionindicative of an enhancement of the hearing perception of the user achieved by the processing of the audio signal due to an increasing amount of feedback suppression and/or better quality of the audio signal when executing the FC algorithm providing for frequency shift and/or phase cancelling in place of the low delay FC management algorithm, which may affect the listening effort and/or comfort and/or naturalness and/or another quality of the sound encoded in the audio signal. Further, the different NC algorithms may differ in dimensionindicative of an adverse effect of the processing of the audio signal by the respective algorithmfor the hearing perception of the user due to an increasing level of artefacts and/or distortions and/or latency when executing the FC algorithm providing for frequency shift and/or phase cancelling in place of the low delay FC management algorithm.

602 604 629 617 616 618 The speech enhancement (SE) algorithm can be configured to provide for an enhancement and/or amplification and/or augmentation of speech contained in the audio signal. In some instances, the SE algorithm may be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver. For instance, the SE algorithm may be applied on the audio signal by audio processing modulewhen classifierattributes at least one class such as speech, speech in quiet, speech in babble, speech in noise, speech from the user, speech from a significant other, background speech, speech from multiple sources, speech in a car, speech in traffic, speech in a reverberating environment, speech in wind noise, speech in a lounge to the audio signal. A corresponding signal processing goal of the enhancement and/or amplification and/or augmentation of speech contained in the input audio signal may thus be predicted based on a class attributed to the audio signal. The SE algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

Some examples of different SE algorithms may comprise a low delay soft SE algorithm, which may provide for an enhancement of soft speech content in the audio signal, a soft SE algorithm, which may provide for an enhancement of soft speech content in the audio signal in a more sophisticated manner, and/or a general SE algorithm, which may provide for an enhancement of general speech content in the audio signal. In particular, the SE algorithm, e.g., the soft SE algorithm and/or the general SE algorithm, may be provided as a non-spatial SE algorithm, which may be configured to provide for speech enhancement in a non-directional manner, and/or as a spatial SE algorithm, which may be configured to provide for speech enhancement in a directional manner.

502 504 502 505 110 120 210 503 504 505 The different SE algorithms may thus be each associated with a performance index differing in at least one of dimensions-. To illustrate, the different SE algorithms may differ in dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,due to an increasing power consumption and/or computational load and/or memory requirement and/or communication bandwidth when executing the soft SE enhancement algorithm and/or the general SE algorithm in place of the low delay soft SE enhancement algorithm. Further, the different SE algorithms may differ in dimensionindicative of an enhancement of the hearing perception of the user achieved by the processing of the audio signal due to an increasing amount and/or quality of the speech enhancement when executing the soft SE enhancement algorithm and/or the general SE algorithm in place of the low delay soft SE enhancement algorithm, which may affect the listening effort and/or comfort and/or naturalness and/or another quality of the sound encoded in the audio signal. Further, the different SE algorithms may differ in dimensionindicative of an adverse effect of the processing of the audio signal by the respective algorithmfor the hearing perception of the user due to an increasing level of artefacts and/or distortions and/or latency when executing the soft SE enhancement algorithm and/or the general SE algorithm in place of the low delay soft SE enhancement algorithm.

629 602 604 616 618 The impulse noise cancelling (INC) algorithm may be configured to determine a presence of an impulse in the input audio signal and to reduce a signal level of the audio signal at the impulse, e.g., to reduce an occurrence of sudden loud sounds in the audio signal, which may be caused, e.g., by a shock on the hearing device, or wherein the signal may be kept at a level such that the sound remains audible by the user and/or, when an occurrence of speech is determined at the impulse, the signal level is not reduced. For instance, the INC algorithm may be executed by default by audio processing module, e.g., to account for a previously known signal processing goal to compensate for any impulse noise which may be present in the audio signal. The INC algorithm may also be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver, e.g., when an impulse has been determined to be contained in the audio signal. The INC algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

502 504 502 505 110 120 210 503 504 505 Some examples of different INC algorithms may comprise a low delay INC algorithm, which may provide for a basic impulse noise cancelling, and an INC algorithm, which may provide for impulse noise cancelling in a more sophisticated manner. The different INC algorithms may thus be each associated with a performance index differing in at least one of dimensions-. To illustrate, the different INC algorithms may differ in dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,due to an increasing power consumption and/or computational load and/or memory requirement and/or communication bandwidth when executing the INC algorithm in place of the low delay INC algorithm. Further, the different INC algorithms may differ in dimensionindicative of an enhancement of the hearing perception of the user achieved by the processing of the audio signal due to a decreased presence of impulse noise and/or improved quality of the audio signal when executing the INC algorithm in place of the low delay INC algorithm, which may affect the listening effort and/or comfort and/or naturalness and/or another quality of the sound encoded in the audio signal. Further, the different INC algorithms may differ in dimensionindicative of an adverse effect of the processing of the audio signal by the respective algorithmfor the hearing perception of the user due to an increasing level of artefacts and/or distortions and/or latency when executing the INC algorithm in place of the low delay INC algorithm.

616 602 The acoustic object separation (AOS) algorithm can be configured to separate audio content representative of sound emitted by at least one acoustic object from the input audio signal. More recently, one or more neural networks (NNs) have been employed to provide such a separation of sound emanated from one or more specific acoustic objects. In this regard, the AOS algorithm may be configured to separate the sound emanated from such an acoustic object by at least one deep neural network (DNN). In particular, the AOS algorithm may comprise an acoustic object separator configured to separate sound generated by different acoustic objects, for instance an own voice of the user, a conversation partner, passengers passing by the user, vehicles moving in the vicinity of the user such as cars, airborne traffic such as a helicopter, a sound scene in a restaurant, a sound scene including road traffic, a sound scene during public transport, a sound scene in a home environment, and/or the like. Examples of such an acoustic object separator are disclosed in international patent application Nos. PCT/EP 2020/051 734 and PCT/EP 2020/051 735, and in German patent application No. DE 2019 206 743.3. The separated audio content generated by the different acoustic objects can then be further processed, e.g., by emphasizing the audio content generated by one acoustic object relative to the audio content generated by another acoustic object and/or by suppressing the audio content generated by another acoustic object. For instance, separating an own voice of the user from the input audio signal may be employed different applications, e.g., a phone call and/or a steering of the hearing device and/or hearing system. E.g., a user command received via user interfacemay include an input audio signal provided by input transducer. Separating the user's own voice from the input audio signal may then be employed to extract the user command from the input audio signal.

617 616 618 A corresponding signal processing goal of the audio content separation and/or emphasizing or suppressing dedicated acoustic objects in the input audio signal may be predicted based on the audio signal, e.g., depending on classifierattributing at least one corresponding class to the audio signal, wherein such a classifier may be also be implemented by the acoustic object separator of the AOS algorithm. The AOS algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

115 125 502 110 115 125 502 120 310 602 604 629 617 616 618 The binaural synchronization (BS) algorithm can be configured to provide for a synchronization between an audio signal received from input transducer,,in first hearing deviceand from input transducer,,in second hearing deviceof hearing system, e.g., with regard to binaural cues indicative of a difference of a sound detected on a left and a right car of the user. In some instances, the BS algorithm may be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver. For instance, the BS algorithm may be applied on the audio signal by audio processing modulewhen classifierattributes at least one class such as speech, nonspeech, speech in quiet, speech in babble, speech in noise, speech from the user, speech from a significant other, background speech, speech from multiple sources, speech in a car, speech in traffic, speech in a reverberating environment, speech in wind noise, speech in a lounge, music and/or the like to the input audio signal. A corresponding signal processing goal of the synchronization between an input audio signals may thus be predicted based on a class attributed to the audio signal. The BS algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

115 125 502 602 604 629 617 616 618 The beamforming (BF) algorithm can be configured to provide for a beamforming of audio content in the audio signal, e.g., with regard to a location of an acoustic object in the environment of the user and/or with regard to a direction of arrival (DOA) of sound detected by input transducer,,and/or with regard to a directivity of the acoustic beam in a front and/or back direction of the user. In some instances, the BF algorithm may be applied depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver. For instance, the BF algorithm may be applied on the audio signal by audio processing modulewhen classifierattributes at least one class such as speech, nonspeech, speech in quiet, speech in babble, speech in noise, speech from the user, speech from a significant other, background speech, speech from multiple sources, speech in a car, speech in traffic, speech in a reverberating environment, speech in wind noise, speech in a lounge to the audio signal. A corresponding signal processing goal of the enhancement and/or amplification and/or augmentation of speech contained in the input audio signal may thus be predicted based on a class attributed to the audio signal. The BF algorithm may also be applied depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

115 110 125 120 Some examples of different BF algorithms may comprise a low delay BF algorithm, which may provide for a directivity of an acoustic beam in a front direction of the user and/or a directivity in a rear direction of the user, a monaural BF algorithm, which may be adaptive, e.g., with regard a directivity and/or width of the acoustic beam, or static, and/or a guided BF algorithm, which may be configured to guide the beam to a side and/or back of the user, and/or a binaural BF algorithm, which may employ an audio signal received from input transducerof first hearing deviceand an audio signal received from input transducerof second hearing device.

502 505 110 120 210 503 504 505 To illustrate, the different BF algorithms may differ in dimensionindicative of an impact of the respective audio processing algorithmon resources available in hearing device,,due to an increasing power consumption and/or computational load and/or memory requirement and/or communication bandwidth when executing the monaural BF algorithm and/or the binaural BF algorithm in place of the low delay BF algorithm and/or when executing the monaural adaptive BF algorithm and/or the binaural BF algorithm in place of the monaural static BF algorithm and/or when executing the binaural BF algorithm in place of the monaural BF algorithm. Further, the different BF algorithms may differ in dimensionindicative of an enhancement of the hearing perception of the user achieved by the processing of the audio signal due to an improved beamforming and/or quality of the audio signal when executing the monaural BF algorithm and/or the binaural BF algorithm in place of the low delay BF algorithm and/or when executing the monaural adaptive BF algorithm and/or the binaural BF algorithm in place of the monaural static BF algorithm and/or when executing the binaural BF algorithm in place of the monaural BF algorithm, which may affect the listening effort and/or comfort and/or naturalness and/or another quality of the sound encoded in the audio signal. Further, the different SE algorithms may differ in dimensionindicative of an adverse effect of the processing of the audio signal by the respective algorithmfor the hearing perception of the user due to an increasing level of artefacts and/or distortions and/or latency when executing the monaural BF algorithm and/or the binaural BF algorithm in place of the low delay BF algorithm and/or when executing the monaural adaptive BF algorithm and/or the binaural BF algorithm in place of the monaural static BF algorithm and/or when executing the binaural BF algorithm in place of the monaural BF algorithm.

6 7 FIGS.and 505 631 632 633 634 505 505 631 634 505 631 634 502 504 631 634 505 631 634 505 631 634 505 505 631 634 631 634 631 634 505 505 631 634 505 631 634 631 634 In some implementations, as illustrated in, the different audio processing algorithmsmay be grouped in different sets,,,of audio processing algorithms. In some instances, at least one of the audio processing algorithmsincluded in a first set-and at least one of the audio processing algorithmsincluded in a second set-different from the first set are configured to provide for the same signal processing goal and/or are associated with a performance index differing in at least one of dimensions-. In some instances, at least one sets-comprises at least two different audio processing algorithms. In some instances, at least two sets-share at least one of the audio processing algorithms. Thus, an intersection of the at least two sets-may include the at least one of audio processing algorithm. At least one of audio processing algorithmsmay thus be included in the first set-and in the second set-. In some instances, at least one of sets-includes at least one of the audio processing algorithmsdifferent from the audio processing algorithmsincluded in the remaining sets-. Thus, the at least one of the audio processing algorithmincluded in the at least one set-may be excluded from a union of the remaining sets-.

631 634 631 634 631 634 505 631 634 631 634 631 634 505 631 634 631 634 631 634 505 631 634 In some instances, a performance index indicative of a performance of the audio processing algorithms included in sets-may be associated with each of sets-. The performance index associated with sets-may have at least one dimension, which may correspond the at least one dimension of the performance index associated with the audio processing algorithmsincluded in the set-. For instance, the first set-and the second set-may each be associated with a performance index indicative of the performance index associated with each of audio processing algorithmsincluded in the set-. As another example, the first set-and the second set-may each be associated with a performance index indicative of the largest and/or smallest performance index associated with the audio processing algorithmsincluded in the set-.

505 631 634 629 631 634 505 631 634 629 631 634 631 634 505 505 631 634 631 634 505 505 505 629 631 634 505 602 604 505 617 505 616 618 In some instances, the different audio processing algorithmsgrouped in different sets-may correspond two different operational modes of audio processing module, wherein, in each operational mode-, one or more of the audio processing algorithmsincluded in the respective set-may be applied to the audio signal. Thus, when audio processing moduleis operating in one of operational modes-corresponding to one of sets-including the at least one audio processing algorithm, the audio processing algorithmincluded in set-can be applied on the audio signal, and when the operational mode-includes at least two audio processing algorithms, at least one of the audio processing algorithmscan be applied on the audio signal and/or the at least two audio processing algorithmscan be applied in a sequence and/or in parallel to the audio signal. In some instances, when audio processing moduleis operating in one of operational modes-, the number and/or type of audio processing algorithmsapplied on the audio signal may be determined depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiver. For instance, the number and/or type of audio processing algorithmsapplied on the audio signal may be determined depending on at least one class attributed to the audio signal by classifier. The number and/or type of audio processing algorithmsapplied on the audio signal may also be determined depending on a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit.

6 7 FIGS.and 631 634 631 631 505 631 110 120 210 505 631 502 503 504 As illustrated in, the different operational modes-may comprise, for example, a first mode, which may be denoted by a low complexity and most natural. Thus, an audio processing provided in first modeby at least one audio processing algorithmincluded in first setmay provide for a rather natural sound perception and may also be rather conservative regarding resources available in hearing device,,required to execute the audio processing algorithm. Accordingly, first modemay be associated with a performance index having dimensionindicative of an impact on available resources being rather high, e.g., due to a rather small footprint of the audio processing on the available resources, and/or dimensionindicative of an enhancement of the hearing perception being rather low, e.g., due to a rather small modification of the audio signal provided by the audio processing, and/or dimensionindicative of an adverse effect of the audio processing being rather high, e.g., due to a rather small delay and/or sound distortion caused by the audio processing.

632 632 505 632 631 110 120 210 632 502 631 503 631 504 631 A second modemay be denoted by a medium complexity, still natural, and more performant. Thus, an audio processing provided in second modeby at least one audio processing algorithmincluded in second setmay provide for a less natural sound perception as compared to first mode, may be less conservative regarding required resources available in hearing device,,, and may be prone to deliver a larger impact on the audio signal. Accordingly, second modemay be associated with a performance index having dimensionindicative of an impact on available resources being reduced as compared to the performance index associated with first mode, but may still be rather high, e.g., due to an increased footprint of the audio processing on the available resources, and/or dimensionindicative of an enhancement of the hearing perception being increased as compared to the performance index associated with first mode, e.g., due to an enhanced modification of the audio signal provided by the audio processing, and/or dimensionindicative of an adverse effect of the audio processing reduced as compared to the performance index associated with first mode, e.g., due to an increased delay and/or sound distortion caused by the audio processing.

633 633 505 633 632 110 120 210 633 502 632 503 632 504 632 A third modemay be denoted by a maximum complexity, less natural, and most performant. Thus, an audio processing provided in third modeby at least one audio processing algorithmincluded in third setmay provide for a rather unnatural or artificial sound perception as compared to second mode, may be rather expensive regarding required resources available in hearing device,,, and may accomplish an even larger impact on the audio signal. Accordingly, third modemay be associated with a performance index having dimensionindicative of an impact on available resources rather small as compared to the performance index associated with second mode, e.g., due to a maximum footprint of the audio processing on the available resources, and/or dimensionindicative of an enhancement of the hearing perception being rather high as compared to the performance index associated with second mode, e.g., due to an extensive modification of the audio signal provided by the audio processing, and/or dimensionindicative of an adverse effect of the audio processing further reduced as compared to the performance index associated with second mode, e.g., due to an even larger delay and/or sound distortion caused by the audio processing.

634 634 632 634 633 633 634 633 634 616 602 604 618 633 634 617 As illustrated, a fourth modemay also be denoted by a maximum complexity, less natural, and most performant. Thus, a performance index associated with fourth modemay have similar values in the at least one dimension-as compared to third mode. In particular, a tradeoff between operating in third modeand fourth modemay then be similar. Accordingly, an operation in third modeor fourth modemay be controlled depending on a preference of the user, e.g., based on a user command provided by user interface, and/or depending on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiverand/or depending on sensor data, which may be provided by sensor unit. E.g., a decision between operating in third modeor fourth modemay be based on a class attributed to the audio signal and/or a class attributed to the sensor data by classifier.

7 FIG. 631 634 505 631 634 751 752 110 120 210 751 752 631 634 632 633 634 771 772 773 774 775 776 110 120 210 771 776 751 752 629 632 634 632 633 761 763 762 110 120 210 761 763 751 752 629 632 633 schematically illustrates a Venn diagram of the exemplary sets-of exemplary audio processing algorithms. An intersection of all four sets-comprises an expansion algorithmand a maximum power output (MPO) algorithm. In this way, a core functionality of hearing device,,may be provided by executing algorithmand/or algorithmin parallel and/or in sequence in each operational mode-. An intersection of second set, third set, and fourth setadditionally comprises a soft SE algorithm, an FC algorithm providing for frequency shift and/or phase cancelling, a FreqC and/or frequency lowering algorithm, an advanced CR/gain algorithm, an advanced INC algorithm, and a WNC algorithm. Accordingly, at least one additional functionality of hearing device,,may be provided by also executing at least one of algorithms-in parallel and/or in sequence, e.g., with algorithm,, when audio processing moduleoperates in any of operational modes-. An intersection of second setand third setadditionally comprises a traditional/hybrid NC algorithmproviding for a non-spatial noise cancelling, a directional NC algorithmproviding for a noise cancelling in a specific direction or location relative to the user, and a general SE algorithmproviding for an enhancement of speech content. Accordingly, at least one additional functionality of hearing device,,may be provided by also executing at least one of algorithms-in parallel and/or in sequence, e.g., with algorithm,, when audio processing moduleoperates in any of operational modes,.

631 711 712 713 714 715 716 110 120 210 713 716 751 752 629 631 632 721 722 110 120 210 721 722 751 752 629 632 633 731 110 120 210 731 751 752 629 633 634 743 741 742 110 120 210 741 743 751 752 629 634 Furthermore, first setexclusively includes a low delay BF algorithm, which may direct the beam to the front of the user, a low delay soft SE algorithm, a low delay NC algorithmproviding for noise cancelling in a non-spatially resolved manner, a low delay INC algorithm, a low delay CR/gain algorithm, and a low delay FB management algorithm. Accordingly, at least one additional functionality of hearing device,,may be provided by also executing at least one of algorithms-in parallel and/or in sequence, e.g., with algorithm,, when audio processing moduleoperates in first operational mode. Second setexclusively includes a monaural BF algorithm, which may be adaptive, e.g., with regard a directivity and/or width of the acoustic beam, and a guided BF algorithm, which may be configured to guide the beam to a side and/or back of the user. Accordingly, at least one additional functionality of hearing device,,may be provided by also executing at least one of algorithms,in parallel and/or in sequence, e.g., with algorithm,, when audio processing moduleoperates in second operational mode. Third setexclusively includes a binaural BF algorithm. Accordingly, another additional functionality of hearing device,,may be provided by also executing algorithmin parallel and/or in sequence, e.g., with algorithm,, when audio processing moduleoperates in third operational mode. Fourth setexclusively includes a monaural BF algorithm, which may be static, an AOS algorithmincluding at least one DNN for separating audio content of at least one acoustic object in the audio signal, and a NC algorithmimplemented by a DNN, which may provide for noise cancelling in a non-spatial manner. Accordingly, at least one additional functionality of hearing device,,may be provided by also executing any of algorithms-in parallel and/or in sequence, e.g., with algorithm,, when audio processing moduleoperates in fourth operational mode.

713 716 721 722 731 741 743 751 752 761 763 771 776 631 634 713 716 721 722 731 741 743 751 752 761 763 771 776 631 634 602 604 618 616 617 713 716 721 722 731 741 743 751 752 761 763 771 776 634 741 742 634 741 742 743 632 721 722 As noted above, one or more of algorithms-,,,,-,,,-,-may be executed in parallel and/or in sequence in each of operational modes-. A decision, which of algorithms-,,,,-,,,-,-are executed in each of operational modes-may further depend on the audio signal provided by input transducerand/or the audio signal provided by audio signal receiverand/or sensor data, which may be provided by sensor unit, and/or a user command, which may be provided by user interface. E.g., the decision may be based on whether classifierattributes at least one predetermined class to the audio signal and/or sensor data. Additionally or alternatively, selection rules may be applied which may define which of algorithms-,,,,-,,,-,-may be applied at the expense of another and/or in conjunction with one another. To illustrate, when operating in fourth operational mode, the selection rules may define that only one of AOS algorithmor NC algorithm, which may both be implemented as a DNN, may be executed, e.g., in order not to overload the available processing resources. To illustrate, when operating in fourth operational mode, the selection rules may define that each of the AOS algorithmand NC algorithm, when executed, will be executed in conjunction with monaural BF algorithm, e.g., to provide for a good quality of the audio signal. As another example, when operating in second operational mode, the selection rules may define that only one of monaural BF algorithmor guided BF algorithmmay be executed, e.g., in order not to negatively influence a desired effect of the beamforming by forming too many beams.

6 FIG. 623 505 713 716 721 722 731 741 743 751 752 761 763 771 776 502 504 502 504 502 504 502 504 502 504 502 504 502 504 502 504 502 504 Turning back to, target index determination moduleis configured to determine a target index indicative of a target performance of said processing of the audio signal. E.g., the target index may be indicative of an applicability of processing algorithms,-,,,,-,,,-,-relative to the performance index. E.g., the target index may thus constrain an applicability of the processing algorithms and may then also be referred to as a constraint index. E.g., the target index may be indicative of one or more goals of the processing of the audio signal. In particular, the target index can be determined in at least one of dimensions-of the performance index of at least one of the audio processing algorithms. E.g., when the audio processing algorithms are associated with a performance index in three dimensions-, the target index may be determined as a scalar corresponding to one of dimensions-, a pair of values, e.g., a two-dimensional vector, corresponding to two of dimensions-, or a triple, e.g., a three-dimensional vector, corresponding to three of dimensions-. The target index can thus be comparable with at least one dimension-of the performance index associated with a respective audio processing algorithm. E.g., the target index may be determined as a threshold, wherein an audio procession algorithm associated with a performance index in one or more dimensions-exceeding the threshold may be applied to the audio signal, and an audio procession algorithm associated with a performance index in one or more dimensions-falling below the threshold may not be applied to the audio signal. As another example, when the performance index and/or target index are provided as an n-tuple or vector in at least two of dimensions-, the performance index and target index may be compared by determining an absolute value of the n-tuple or vector before the comparison. E.g., the absolute value of the target index may then constitute a threshold for the absolute value of the performance index.

623 602 603 604 605 110 210 614 623 616 618 623 As illustrated, target index determination modulemay also be configured to receive the audio signal provided by input transducerafter it has been converted into a digital signal by analog-to-digital converter, and/or the audio signal provided by audio signal receiver, after it has been decoded by decoder. Further, the information about resources currently available in hearing device,, as acquired by hearing device management module, may also be received by target index determination module. Further, a user command, which may be provided by user interface, and/or sensor data, which may be provided by sensor unit, may also be received by target index determination module. Accordingly, the target index may be determined based on at least one of the user command indicative of the target index; the evaluated audio signal, wherein the target index is determined based on the evaluated audio signal; the evaluated sensor data; and the acquired information about resources available in the hearing device.

304 303 133 135 130 132 136 303 110 210 112 122 302 To illustrate, when the user command indicates that the user is rather allergic to adverse effects of the audio signal processing, e.g., the user dislikes latency and/or artefacts in the reproduced sound, dimensionof the target index may be determined to be rather low. When the evaluated audio signal would indicate a rather simple acoustic environment, e.g., a non-speech scenario and/or little noise scenario, which may require little processing effort, dimensionof the target index indicative of an enhancement of the hearing perception of the user by the audio processing may be determined to be rather low. In particular, shooting at sparrows with guns may thus be avoided. When the sensor data would indicate circumstances requiring a rather elevated enhancement of the hearing perception of the user, e.g., when physiological sensor data provided by physiological sensor-would indicate a medical urgency situation and/or when environmental sensor data provided by environmental sensor-and/or movement sensor data provided by movement sensorwould indicate the user being in a situation of high traffic or another situation requiring high concentration, dimensionof the target index may be determined to be rather high. When the information about resources currently available in hearing device,would indicate the resources being rather low, e.g., a remaining battery power being rather small and/or processor,being rather overloaded with processing tasks, dimensionof the target index indicative of an impact of the audio processing algorithm on available resources available may be determined to be rather low.

625 505 713 716 721 722 731 741 743 751 752 761 763 771 776 629 623 625 625 625 625 625 Audio processing algorithm selection moduleis configured to select, depending on the target index, at least one of processing algorithms,-,,,,-,,,-,-to be applied on the audio signal by audio processing module. To this end, the target index determined by target index determination modulemay be received by audio processing algorithm selection module. Further, the performance index associated with the different audio processing algorithms is available by audio processing algorithm selection module. Audio processing algorithm selection modulecan thus be configured to compare the target index with the performance index associated with the different audio processing algorithms. For instance, as described above, the target index may be provided as a threshold, wherein an audio processing algorithm with a performance index exceeding the threshold may be selected and an audio processing algorithm with a performance index falling below the threshold may be rejected by audio processing algorithm selection module. In some examples, e.g., when multiple audio processing algorithms with a similar or identical processing goal would exceed the threshold, the audio processing algorithm closest to the threshold may be selected by audio processing algorithm selection module.

625 602 603 604 605 618 505 713 716 721 722 731 741 743 751 752 761 763 771 776 617 617 625 617 As illustrated, audio processing algorithm selection modulemay also be configured to receive the audio signal provided by input transducerafter it has been converted into a digital signal by analog-to-digital converter, and/or the audio signal provided by audio signal receiver, after it has been decoded by decoder, and/or sensor data provided by sensor unit. The selection of one or more audio processing algorithms,-,,,,-,,,-,-may then also be based on the audio signal and/or the sensor data. In some instances, the audio signal and/or the sensor data may be classified by classifier. At least one class attributed to the audio signal and/or sensor data by classifiermay then be received by audio processing algorithm selection module. The selection of one or more audio processing algorithms may then also be based on the class attributed to the audio signal and/or sensor data by classifier.

625 631 634 505 631 634 625 505 713 716 721 722 731 741 743 751 752 761 763 771 776 In some implementations, audio processing algorithm selection modulecan select, depending on the target index, one of sets-of audio processing algorithms, and then select, based on the audio signal and/or the sensor data and/or the at least one class associated with the audio signal and/or the sensor data, one or more audio processing algorithms included in the selected set-. In some implementations, audio processing algorithm selection modulecan directly select, depending on the target index, one or more of audio processing algorithms,-,,,,-,,,-,-. From this first selection of the audio processing algorithms, one or more audio processing algorithms may be selected again in a second selection based on the audio signal and/or the sensor data and/or the at least one class associated with the audio signal and/or the sensor data.

623 625 505 713 716 721 722 731 741 743 751 752 761 763 771 776 623 625 502 504 In some implementations, target index determination moduleand/or audio processing algorithm selection modulecan be configured to learn and/or estimate the target index in order to select one or more of audio processing algorithms,-,,,,-,,,-,-mostly adapted to a current scene and/or the user's preferences. For instance, the target index determination moduleand/or audio processing algorithm selection modulemay be implemented as a machine learning (ML) algorithm, e.g., a NN or a DNN. The ML algorithm may be trained by a set of training data comprising previously acquired user commands and/or audio signals and/or sensor data and/or information about resources available in the hearing device, which may be labelled by a corresponding target index having at least one of dimensions-. The trained ML algorithm may then determine the target index and/or select, depending on the target index, one or more of the audio processing algorithms based on input data comprising at least one of a currently received audio signal and/or sensor data and/or information about resources available in the hearing device. Thus, the trained ML algorithm may be configured, e.g., to predict the target index and/or select one or more of the audio processing algorithms without requiring further input from the user, e.g., via a user command.

8 FIG. 801 801 821 811 831 833 834 811 812 117 127 614 821 822 822 821 831 833 834 821 831 821 833 821 835 831 833 834 832 834 836 831 833 834 832 834 836 821 832 834 836 832 834 836 812 831 832 833 834 835 836 schematically illustrates modular audio signal processing algorithmswhich may be implemented as a DNN. As illustrated, audio signal processing algorithmscomprise an encoder partconfigured to encode an input audio signal, and one or more decoder parts,,configured to decode the encoded audio signal. The decoded audio signalmay then be provided, as the processed audio signal, to output transducer,,to output an output audio signal so as to stimulate the user's hearing. As illustrated, encoder partcomprises a plurality of layers, e.g., at least one input layer and a number of hidden layers. The output of the last hidden layerof encoder partcan then be fed into one of decoder parts,,. Thus, a first NN may be provided by encoder partand first decoder part, a second NN may be provided by encoder partand second decoder part, and a third NN may be provided by encoder partand third decoder part. Decoder parts,,are distinguished by a different number of layers,,. E.g., each decoder part,,may comprise an input layer,,receiving the output of encoder part, a number of hidden layers,,, and an output layer,,to output the decoded audio signal. In particular, first decoder partmay comprise a smallest number of layers, second decoder partmay comprise a larger number of layers, and third decoder partmay comprise a largest number of layers.

811 821 835 811 821 834 811 821 831 811 821 834 821 831 502 110 120 210 821 833 502 821 835 502 Thus, a processing of audio signalperformed by third NN,may be more processing intensive as compared to a processing of audio signalperformed by second NN,. Similarly, a processing of audio signalperformed by first NN,may be less processing intensive as compared to the processing of audio signalperformed by second NN,. Accordingly, a performance index associated with first NN,may be larger in dimensionindicative of an impact of the audio processing algorithm on resources available in hearing device,,as compared to a performance index associated with second NN,in dimension, which may be larger than a performance index associated with third NN,in dimension.

811 821 835 811 821 833 811 821 831 811 821 833 821 831 503 821 833 503 821 835 503 Further, a processing of audio signalperformed by third NN,may have a larger impact on the audio signal with regard to an enhancement of the hearing perception of the user as compared to a processing of audio signalperformed by second NN,. Similarly, a processing of audio signalperformed by first NN,may have a smaller impact on the audio signal with regard to an enhancement of the hearing perception of the user as compared to the processing of audio signalperformed by second NN,. Accordingly, a performance index associated with first NN,may be smaller in dimensionindicative of an enhancement of the hearing perception of the user by the processing of the audio signal as compared to a performance index associated with second NN,in dimension, which may be smaller than a performance index associated with third NN,in dimension.

811 821 835 811 821 833 811 821 831 811 821 833 821 831 504 821 833 504 821 835 504 Further, a processing of audio signalperformed by third NN,may also be more prone to have an adverse effect, e.g., with regard to a latency produced by the processing, as compared to a processing of audio signalperformed by second NN,. Similarly, a processing of audio signalperformed by first NN,may be less prone to have an adverse effect as compared to the processing of audio signalperformed by second NN,. Accordingly, a performance index associated with first NN,may be larger in dimensionindicative of the adverse effect as compared to a performance index associated with second NN,in dimension, which may be larger than a performance index associated with third NN,in dimension.

623 502 504 503 625 821 835 821 833 821 831 821 835 821 833 623 502 504 503 625 821 831 821 833 821 835 821 831 821 833 Accordingly, when a target index determined by target index determination modulewould indicate a smaller value in dimensions,, and/or a larger value in dimension, audio processing algorithm selection modulecan be configured to select third NN,or second NN,in place of first NN,, and/or third NN,in place of second NN,. Conversely, when a target index determined by target index determination modulewould indicate a larger value in dimensions,, and/or a smaller value in dimension, audio processing algorithm selection modulecan be configured to select first NN,or second NN,in place of third NN,, and/or first NN,in place of second NN,.

623 502 504 625 502 504 110 120 210 614 616 502 504 503 625 821 831 821 833 821 835 821 831 821 833 As another example, when a target index determined by target index determination modulewould indicate a smaller value in all dimensions-, audio processing algorithm selection modulecan be configured to determine a preference between dimensions-. E.g., when resources available in hearing device,,are rather sparse, which may be indicated by the information acquired by hearing device management module, and/or when the user has a distaste against adverse effects produced by the processing, which may be indicated by the user command provided by user interface, dimension,may be associated with a larger priority as compared to dimension. Accordingly, audio processing algorithm selection modulecan then be configured to select first NN,or second NN,in place of third NN,, and/or first NN,in place of second NN,.

625 502 504 821 831 834 502 504 502 504 625 821 831 833 835 502 504 As another example, processing algorithm selection modulecan be configured to select one or more audio processing algorithms associated with a performance index having a best match with the target index in at least one of dimensions-. E.g., when all available audio processing algorithms,-would be associated with a performance index which is rather far off the target index in at least one of dimensions-, the remaining dimensions-may be associated with a larger priority. Accordingly, audio processing algorithm selection modulecan then be configured to select the best matching NN,,,in the remaining dimensions-.

625 502 504 625 821 831 833 835 502 504 821 831 833 835 As a further example, processing algorithm selection modulecan be configured to select one or more audio processing algorithms associated with a performance index closest to the target index in at least one of dimensions-in which the performance index of all available audio processing algorithms is most distant from the target index. Accordingly, audio processing algorithm selection modulecan then be configured to select the NN,,,closest to the target index in dimension-most distant from the performance index of all available NNs,,,.

821 831 833 835 821 831 833 835 742 634 741 743 751 752 771 776 821 831 833 835 821 831 833 835 741 634 741 743 751 752 771 776 In some implementations, NNs,,,can be configured as a noise cancelling (NC) algorithm. E.g., NNs,,,may be implemented as DNN denoising algorithm, which may be included in setof audio processing algorithms-,,,-. E.g., an audio processing algorithm as disclosed in European application number EP 23164336.2 may be implemented in such a way. In some implementations, NNs,,,can be configured as an acoustic object separation (AOS) algorithm. E.g., NNs,,,may be implemented as DNN separation algorithm, which may be included in setof audio processing algorithms-,,,-. E.g., an audio processing algorithm as disclosed in international patent application Nos. PCT/EP 2020/051 734 and PCT/EP 2020/051 735 may be implemented in such a way.

9 FIG. 311 310 110 210 310 11 811 12 811 13 14 15 811 811 812 812 117 127 illustrates a block flow diagram for an exemplary method of processing input audio signal. The method may be executed by processorof hearing device,and/or another processor communicatively coupled to processor. At operation S, audio signalis received. At operation S, different audio processing algorithms are provided, which are each associated with a performance index indicative of a performance of the respective audio processing algorithm when applied on audio signal. At operation S, a target index relative to the performance index is determined. At operation S, at least one of the processing algorithms is selected depending on the target index. At operation S, audio signalis processed by applying the selected audio processing algorithm on the audio signalto generate a processed audio signal. Subsequently, an output audio signal based on processed audio signalcan be output by output transducer,so as to stimulate the user's hearing.

While the principles of the disclosure have been described above in connection with specific devices and methods, it is to be clearly understood that this description is made only by way of example and not as limitation on the scope of the invention. The above described embodiments are intended to illustrate the principles of the invention, but not to limit the scope of the invention. Various other embodiments and modifications to those embodiments may be made by those skilled in the art without departing from the scope of the present invention that is solely defined by the claims. In the claims, the word “comprising” does not exclude other elements or steps, and the indefinite article “a” or “an” does not exclude a plurality. A single processor or controller or other unit may fulfil the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope.